Nvidia Tips New Volta Architecture for Supercomputer GPUs

SAN JOSE, Calif.—Nvidia refreshed its lineup of GPUs for deep learning and bogus intelligence applications on Wed with the new 5,120-core, 7.5 teraflop Tesla V100 Volta.

The new processor is role of Nvidia's quest to come with a new manner to consistently meliorate calculating performance in the backwash of Moore'south Police force, which many industry leaders agree is pretty much dead. Instead of boosting processor speeds or cramming more transistors onto already-crowded silicon, Nvidia is championing GPU-accelerated computing, which the visitor'due south CEO Jensen Huang (pictured in a higher place) said can offering a 150 pct performance boost every yr.

The Tesla V100, with a make-new architecture called "Volta," represents that latest boost. Information technology's the "adjacent behemothic leap into the new globe" of AI and high-performance computing, Huang said at Nvidia's developers conference here. The V100 will start shipping by the cease of the twelvemonth to data centers owned past Amazon, Microsoft, and other cloud computing providers in several different configurations.

Nvidia Tesla V100

The nigh common off-the-shelf version of the V100 is a $149,000 supercomputer chosen the DGX-ane, which contains 8 V100 processors. There'south also a $69,000, liquid-cooled mini supercomputer called the DGX aimed at researchers who aren't using the cloud, which is powered past four V100s.

The DGX-1 made its debut last year with the V100's predecessor, the Tesla P100. Cheers to the Volta architecture, the V100-powered supercomputers offer five times more superlative teraflops improvement over Pascal chips, the current-generation Nvidia GPU compages, and 15 times more than the Maxwell architecture offers.

The V100 will power servers in the Microsoft Azure cloud and Amazon Web Services, which are the 2 titans of cloud offerings for deep learning. Once installed, they'll back up software similar Microsoft's Cerebral Toolkit and Mxnet, an open-source deep learning framework that is Amazon's preferred AI framework. Nvidia's V100 volition likewise probable power Facebook's Caffe2 framework.

"Our strategy is to create the nearly productive platform for deep learning," Jensen said. He offered several demonstrations of the P100'south potential applications, from an artwork algorithm that can merge the color palette of one photo (an orange sunset, for case) with the subject of another (a beach and clouds, for case) into an entirely new paradigm.

Of grade, the chief benefit of hosting AI processing ability in the cloud is that engineers can do pretty much whatever they want with information technology. To that end, Nvidia as well plans to launch its ain GPU Cloud service, currently in an invite-only beta, which allows anyone to upload their auto algorithms to Volta servers.

"One of the best ways to savor deep learning is for someone else to build this incredibly complicated supercomputer on your behalf," Jensen said.